Site News  |   Monitor Keywords  |   Monitor Archive  |   Organizer  |   Account Info  |

High-availability networking with intelligent failover

High-availability networking with intelligent failover description/claims

This application claims the benefit of U.S. Provisional Application No. 60/708,863, filed Aug. 17, 2005. The entire teachings of the above application are incorporated herein by reference.

This invention relates generally to the field of computer networks. More particularly, this invention relates to maintaining high availability in a computer network utilizing link redundancy and failover control.

Computer networks play an ever expanding role in today's economy. As the available number and types of networked resources increase, combined with increases in speed and affordability of communications bandwidth, networks are becoming in some sense indistinguishable from computing systems. One example of such a reliance on computer networking occurs at the enterprise level as demonstrated by networked storage solutions. Rather than providing physical storage at a client system, the enterprise relies on shared storage whereby high-density, network-accessible storage servers are separately managed from client systems. Such a growing reliance on networks to perform even basic computing services, such as storage, imposes increasing demands for high availability. Any network interruptions can range from a mere inconvenience to an intolerable situation for mission critical applications.

In some cases, mechanisms may be put into place to detect an error in a network connection and to notify a network administrator. The administrator may then take action to identify the source or at least the general location of the error and to take corrective action, such as reconfiguring network resources. Unfortunately, such actions take time and result in interruptions to workflow. Such a manual intensive approach would be hard pressed to meet the high availability requirement of today's mission critical systems.

In other cases, resources are provided to facilitate a failover to a redundant resource. One such example is illustrated in FIG. 1. A high-availability networked computer system 100 includes one or more servers 102a, 102b, 102c (generally 102), each including multiple network interfaces 104a, 104b, 104c, 104d (generally 104), each interface 104 coupled to a different network switch 106a, 106b, 106c, 106d (generally 106). The servers 102 can represent blade server modules of a blade server system. The multiple network interfaces 104 of each server 102 are controlled by a teaming/failover controller that monitors the internal link status at the physical layer. Upon detecting a link drop, the teaming/failover controller fails over to one of the other network interfaces 104, thereby reestablishing communications over a different internal link and through a different switch 106.

Some systems provide a mechanism to monitor external link state through the physical layer of the external ports (EXTA-EXTD). In response to detecting an external link failure, the mechanism also triggers an internal link drop on all of the corresponding internal ports (INTA-INTD) of the associated switch 106. This link drop initiates the failover mechanism provided on each of the servers 102 with an active link to the effected switch 106 so that it could properly failover, switching its active link to another one of the network interfaces.

Unfortunately, monitoring external link states at the physical layer of an external switch port is susceptible to frequent “link flapping” issues experienced when enabling that port. Only having physical-layer status, the system is unable to distinguish between “real” link failures and the intermittent link flapping events. Consequently, this leads to unnecessary internal link drops and failover events at the respective server 102 causing it to ping-pong between selecting the appropriate active link.

What are needed are methods and systems to provide high availability network connectivity in a computer system. The present invention satisfies these needs and provides additional advantages. In particular, the present invention provides processes and systems for monitoring external link state using state information obtained from an OSI model layer 2 or higher protocol running on the external link. Relying on information from such a protocol as the spanning tree protocol (STP), it is possible to avoid falsely identifying external link failures due to link flapping. In addition, the present invention provides processes and systems providing flexibility in the definition of an external failure event by providing configurable triggers. Only when the trigger event occurs, is an internal link drop initiated causing failover to a redundant link. Thus, a STP state can be monitored on at least one identified Virtual Local Area Network (VLAN). Further, the STP state can be monitored for static trunk groups or LACP (Link Aggregation Control Protocol (LACP) trunk groups.

In one aspect, the invention features a process for maintaining network connectivity in a computing device coupled to a network through at least one spanning-tree-protocol enabled switch. The computing device includes multiple network interfaces adapted in a failover configuration. One of the network interfaces is active and in communication with an internal switch port, such that the active network interface is in switchable communication with a remote network through the switch. External links from the switch to the remote network can use one or more external switch ports, depending upon a trunking configuration. The STP state of the one or more external switch ports is monitored. An external failure event is determined based on the monitored STP states of the one or more external switch ports. Upon determining an external failure event, one or more internal links coupled between the active network interface and the internal switch port are deactivated, or “dropped” in response to the identified external failure event. A failover from the active network interface to another one of the multiple network interfaces is initiated in response to the deactivated internal link.

In another aspect, the invention features a network-enabled computer system for maintaining high availability network connectivity between the computer system and a network. The computer system includes a computing device having multiple network interfaces adapted in a failover configuration. Each network interface is coupled to one side of a respective internal communication link with one of the network interfaces being active. A spanning-tree-protocol enabled switch has an internal port coupled to another side of the respective internal communication link. The active network interface is in switchable communication with at least one external port of the switch coupled to the network through an external communication link. An intelligent failover controller includes a fault monitor in communication with the STP enabled switch for monitoring a STP state at the at least one external port. The intelligent failover controller also includes a link-drop controller in communication with the fault monitor. The link-drop controller selectively initiates a link drop on one or more of the internal communication links in response to the monitored STP state. The active network interface fails over to another one of the multiple network interfaces in response to the link drop.

The above and further advantages of this invention may be better understood by referring to the following description in conjunction with the accompanying drawings, in which like numerals indicate like structural elements and features in the various figures. For clarity, not every element may be labeled in every figure. The drawings are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the invention.

FIG. 1 illustrates a block diagram of an exemplary embodiment of a high-availability, fault-tolerant communications network.

FIG. 2 illustrates a block diagram of one embodiment of a high-availability, fault-tolerant communications network.

• Provisional Patent
• Utility Patent

• What Is a Patent?
• What Is a Trademark or Servicemark?
• What Is a Copyright?
• Patent Laws